Intelligence and Psychological Testing

One ability or many?

• Spearman used factor analysis to argue for a general intelligence factor, g, underlying all specific abilities.
• Gardner proposed multiple intelligences — linguistic, logical-mathematical, spatial, musical, bodily-kinesthetic, interpersonal, intrapersonal, naturalist — arguing they are independent.
• Sternberg's triarchic theory keeps it to three: analytical (the kind school tests), creative, and practical.
• Emotional intelligence — perceiving, understanding, managing and using emotion.

Across the lifespan, fluid intelligence (reasoning about novel problems) declines with age, while crystallised intelligence (accumulated knowledge and vocabulary) holds or rises.

What makes a test worth anything

Three requirements. Standardisation: the test is given identically and scored against a norm group, producing the familiar normal curve with a mean of $100$ and a standard deviation of $15$. Reliability: consistent results, checked by test-retest or split-half. Validity: it measures what it claims — content validity (covers the domain) and predictive validity (forecasts the outcome it should).

The original ratio score was $\text{IQ} = (\text{mental age} / \text{chronological age}) \times 100$, which breaks down for adults, so modern tests such as the WAIS score by comparison with same-age peers instead. Because raw performance has risen across generations — the Flynn effect — tests must be renormed to keep the mean at $100$.

Differences and fairness

High heritability within groups does not explain a gap between groups — the same seed in poor and rich soil gives different harvests. Stereotype threat: reminding people of a negative stereotype about their group lowers their performance on the spot. A test is biased in the technical sense only if it predicts less accurately for one group; that is different from a test being socially unfair. Finally, aptitude tests aim to predict future performance, achievement tests measure what has been learned.